key: cord-0868496-ypsw6odn
authors: Cowan, Sarah K.; Bruce, Tricia C.; Perry, Brea L.; Ritz, Bridget; Perrett, Stuart; Anderson, Elizabeth M.
title: Discordant benevolence: How and why people help others in the face of conflicting values
date: 2022-02-18
journal: Science advances
DOI: 10.1126/sciadv.abj5851
sha: 334bb440f928733d4babd4d63feec8fa2b484d93
doc_id: 868496
cord_uid: ypsw6odn

What happens when a request for help from friends or family members invokes conflicting values? In answering this question, we integrate and extend two literatures: support provision within social networks and moral decision-making. We examine the willingness of Americans who deem abortion immoral to help a close friend or family member seeking one. Using data from the General Social Survey and 74 in-depth interviews from the National Abortion Attitudes Study, we find that a substantial minority of Americans morally opposed to abortion would enact what we call discordant benevolence: providing help when doing so conflicts with personal values. People negotiate discordant benevolence by discriminating among types of help and by exercising commiseration, exemption, or discretion. This endeavor reveals both how personal values affect social support processes and how the nature of interaction shapes outcomes of moral decision-making.

Living in complex, cooperative communities frequently requires benevolence, or actions that strengthen and support the well-being and interests of others (1, 2) . Understanding when and why humans engage in supportive actions is among the most fundamental questions in sociology. Often, decisions about whether and how to respond to requests for help from members of one's personal network are a straightforward matter of material or logistical wherewithal. Sometimes, however, a request for help introduces a moral conflict. For example, being asked to help a friend cheat on an exam is likely to produce conflict between the values of benevolence and integrity. Such situations may generate cognitive and emotional dissonance or discomfort, but existing empirical research provides little guidance about how such conflicts are resolved in everyday life, i.e., outside the context of the laboratory (3).

Our research interrogates how people resolve incompatible values in the context of morally fraught requests for support from close friends and family members. In short, we ask: When at a moral crossroads, how do people manage? In answering this question, we integrate and extend two literatures central to sociology's articulation of what motivates benevolent action: that of support within social networks and that of moral decision-making. Identifying how people respond to moral conflict when close friends and family members are in need reveals, on the one hand, the importance of personal values to social support processes and, on the other hand, how the nature of interaction shapes outcomes of moral decision-making. It responds to calls for a sociological approach to morality that "...asks not only about the sources of the moral codes on which people rely but also about how people balance moral principles with other considerations in making judgments and taking action" (4). We focus not on the mere existence of moral dilemmas or contradictory actions, which are themselves unremarkable. Rather, we strive to capture and characterize, using a mixed-methods strategy, the processes underlying real-world moral decision-making in the context of dynamic relationships between individuals and their social environments (5) . This research holds critical implications for understanding the power of social connection in the face of polarized values.

Empirically, we examine the case of Americans' willingness to help a close friend or family member who has decided to have an abortion. We argue that abortion is an ideal case for interrogating value conflict for three reasons. First, opinions on abortion are more polarized than opinions on most other issues (6, 7) . A large minority of Americans have consistent and strongly held abortion positions, while the rest express moral uncertainty as a result of internal value conflict (8) . Second, abortions are common in the United States (9) . Third, the financial and logistical requirements of obtaining an abortion compel requests for help. To examine this case, we draw upon both large-scale survey data and in-depth interviews in a mixed-methods approach. The survey data come from new questions added to the 2018 General Social Survey (GSS) regarding Americans' attitudes toward the morality of abortion and willingness to help a friend or family member who has decided to have an abortion. The interview data come from 74 in-depth interviews with people living in the United States regarding their opinions on abortion, drawn from the 2019 National Abortion Attitudes Study (NAAS) (study N = 217).

These data illustrate that many Americans morally opposed to abortion are nonetheless willing to provide assistance to persons they know who are seeking an abortion. We label the inclination to provide help that is inconsistent with one's personal values discordant benevolence. The major contributions of this research lie in (i) documenting discordant benevolence in a setting consequential to people's lives and (ii) demonstrating how people justify and make sense of discordant benevolence as a way of resolving moral conflicts. We find that conflicting values are resolved through circumscribed action and enabling strategies including commiseration, exemption, and discretion, providing a critical window into social support and moral decision-making processes within established relationships. While this examination focuses on abortion, we anticipate that these enabling strategies are used at other moral crossroads as well.

The literature on support in social networks establishes when and why people are motivated to extend support but contains little discussion of how personal values might influence social support processes. For this, we turn to the social psychological theory of moral decision-making. We summarize this literature before offering our critique and contribution. According to this theory, when faced with a decision about whether and how to help someone, people weigh the potential consequences of different courses of action for their set of relatively stable values (29) . The degree to which people feel a moral obligation to help depends, in large part, on whether the helping action is likely to promote values that are of highest priority (30, 31) . This theory sees values as standards shaping how people evaluate situations, institutions, other individuals and groups, and oneself. Unlike norms and attitudes, values are organized hierarchically such that some values are more important than others. The relative importance of the values invoked in a given situation guides individual action (31) . Consequently, according to this theory, trade-offs between competing values guide decision-making and behavior across a range of contexts and situations (32, 33) .

Receiving a request for help from a friend or family member activates multiple, potentially competing values (32, 33) . If a family member asks for financial assistance, for example, the value of benevolence may arise alongside the value of security, forcing a decision between providing resources to another or retaining them for one's own financial well-being. In cases of conflict, the cognitive and emotional processes driving behavior are complex, dynamic, and potentially unconscious. The attractiveness of helping depends on the importance of activated values and the extent to which they are perceived as relevant in a given situation and at a particular point in time (31) .

Despite the important contributions of this body of research to our understanding of prosocial behavior, theory and research on moral decision-making emphasize relatively low-stakes intra-individual processes and minimize fraught interactional ones (22) . This is related, in part, to the conditions under which the theory of moral decision-making has been developed and empirically tested. For example, much research in this area focuses on explaining general prosocial behavior [e.g., volunteering or charitable giving (34) ] or social attitudes toward out-groups [e.g., attitudes toward immigration (35) ], where social norms governing interpersonal relationships are less salient and there is no clear moral conflict. Under these conditions, it is reasonable to assume that people make individual, logical decisions based largely on the strength or salience of personal values (36) . When moral decision-making has been empirically tested in interactional contexts, subjects have been offered the choice to help a stranger in highly controlled and artificial laboratory settings where helping decisions have abstract or insignificant consequences (37) (38) (39) . In the real world, however, moral judgments depend heavily on meaning making in situ and on our ideas about social relationships, including who, exactly, is involved (3, 40) . Sociologists have advocated for studies of moral decision-making in more realistic (even if still hypothetical) situations (3, 41) . We offer one such situation.

Along the same lines, research on moral dilemmas has been criticized for assuming a degree of controlled cognition that is inconsistent with how people make decisions in everyday life (42) (43) (44) . According to dual-process models of cognition, people largely act quickly and effortlessly based on cognitively stored values that are situationally applied and only rarely mediated by deliberative thought (45) (46) (47) . In line with this model, sociologists argue that people draw on relationship schemas, or "imagined" relationships, to determine how to act in the face of moral dilemmas (4) . Evaluations are based on gut reactions and emotions more than complex, cognitive consideration of moral principles-in the context of personal relationships, we intuit what is right and what is wrong.

Reliance on implicit schemas may be particularly likely in situations where moral lines are blurry, leading to actions that are seemingly irrational or inconsistent with expectations. Polletta notes that "imagined relationships may allow us to make judgments or take 3 of 16 actions that are moral enough" [emphasis in original, (4) ], in part by relying on emotional associations that just "feel" right. In cases of moral conflict, the decision whether to help a friend or family member in need is affective and intuitive, and a deliberative rationale for action may be constructed post hoc (48, 49) . To behave contrary to one's own values-based sense of moral obligation creates dissonance, or cognitive and emotional discomfort. Such discomfort may engage distinct decision-making processes that reflect moral intuition rather than moral reasoning-processes that may suggest contradictory actions (48) . How moral decision-making operates in supportive exchanges between friends or family members remains an open question.

These ideas expose a critical gap in our understanding of helping behavior. In the complex social environments in which people live, most important decisions, including those regarding helping others, are made in the context of social interactions with individuals embedded in relationships and social networks (50, 51) . We anticipate that personal values may operate differently in decisions to help friends or family members relative to helping strangers or engaging in general prosocial behavior. We also expect that people will make intuitive, emotional decisions in the context of a close relationship that may contradict other moral values (4, 48) . Consequently, we must extend and refine the theory of moral decision-making by integrating insights from the social support literature.

Integrating theories of social support provision in personal networks with those of moral decision-making yields several important insights. First, people feel a moral obligation to help those they know, far more so than strangers or abstract others, the focus of the extant literature. Second, this moral obligation makes moral conflicts more likely. Third, such conflicts will produce dissonance, motivating individuals to behave or explain their actions in a way that preserves, to the degree possible, both sets of values. Fourth, strategies for negotiating dissonance might include changing the action (e.g., type or level of support) or the action perception (e.g., the meaning of the support).

When facing a request for help from a close associate that conflicts with other personally held values, people must resolve competing values. We argue that these situations can give rise to discordant benevolence. Again, we define discordant benevolence as the extension of help that satisfies one's sense of responsibility to participate in supportive exchanges despite conflict with another strongly held value. Alternative responses to requests for help that produce moral conflict may eschew benevolence entirely to hold firm on another personal value. For example, a person could ignore, deny, or divert the request for help or even sever the relational tie; these responses are not discordant and therefore lie outside our theoretical scope conditions.

With regard to our case of abortion in the United States, we first describe abortion-seeking in the United States. We then describe our data and methods of analysis. Turning to the empirics, we initially do not narrow our sample to fulfill our scope conditions. Instead, we consider American adults who hold a range of abortion attitudes and are willing to extend various forms of support, including none at all. We analyze GSS data in which Americans outline their willingness to offer various types of help to a close friend or family member seeking an abortion, dependent upon their stance toward abortion's morality. We show that Americans who perceive abortion as morally wrong offer fewer types of support than Americans who are not morally opposed to abortion. But the majority of those morally opposed to abortion are willing to extend emotional support, and a substantial minority are willing to directly enable the abortion through provision of logistical or financial support. The survey data reveal (i) the importance of moral decision-making within personal networks and (ii) discordant benevolence as a common approach to morally conflicted requests for help.

Next, we focus our analysis on American adults who do engage in discordant benevolence, that is, Americans who are morally opposed to abortion and extend help (or are willing to) in opposition to their values on abortion. To understand how people reason through their engagement in discordant benevolence, we turn to in-depth interviews with 74 Americans who deem abortion morally wrong, drawn from a strategic sampling frame of Americans that spans the abortion attitudinal spectrum. We analyzed all 74 of these transcripts, but report results for only the subset that extended help, i.e., interviewees willing to engage in discordant benevolence.

This article examines how people resolve conflicting values in moral decision-making within personal community networks using the case of abortion in the United States. More than 800,000 abortions were performed in 2017; if current rates persist, a quarter of U.S. women will have had an abortion by age 45 (9) . (We note that people of all genders seek abortions, and we use gender-neutral language when we can. When referencing a study in which the data are only of women, or when interview respondents specifically refer to women, we use female pronouns.) Abortion is an ideal case for examining moral decision-making for three primary reasons, each discussed in turn.

First, attitudes toward abortion reflect value divergence across the U.S. population. Americans are sharply divided when it comes to abortion (6) , and they continue to be so even when other moral issues have become less politicized (7) . Opinion on abortion is one of only a few issues on which many Americans have a clear, stable position (52) . Equal shares of the U.S. population self-identify as "pro-life" (48%) and "pro-choice" (48%), suggesting that the public is largely split on the issue of abortion (53) . Attitudes toward abortion can stand in as implicit statements about deeply held values like traditionalism or self-direction (54, 55) . Views about abortion's legality fall along a continuum, ranging from the belief that abortion should be forbidden under any circumstance to allowable in all circumstances (56, 57) . Most Americans believe abortion should be legal but subject to a varying set of constraints (58) . The inclination to arbitrate by circumstance means some abortion scenarios will raise more value conflict than others.

A second reason why abortion is an ideal case to test moral decisionmaking within personal networks is that getting an abortion frequently necessitates help from others. Given the financial and logistical requirements of undergoing an abortion, few people can do so without help. The abortion procedure is expensive, costing an average of about $500 during the first trimester (59, 60) . For low-income women in particular, who constitute three-quarters of abortion patients (61) , securing funds for an abortion can be especially challenging given widespread laws against Medicaid and private insurance coverage for abortion (62) (63) (64) . Procedural costs, paired with limited options for financial assistance, increase the likelihood that persons seeking an abortion will ask a friend or family member to help pay for it.

Beyond the expense of the procedure itself, getting an abortion can introduce numerous ancillary costs. Patients often need to drive long distances to access an abortion clinic (65) . Persons who do not live within a reasonable distance of an abortion clinic must also take time off from work and find a place to stay. Many states impose waiting periods ranging from 18 to 72 hours (61, 66) . Nearly 60% of abortion patients are parents, most of whom require childcare during and after the procedure (67) . To overcome these financial and logistical barriers, abortion patients frequently turn to charitable organizations, family, and friends (60) . Abortion seekers are careful to avoid stigma (68, 69) , but likely need to engage personal networks to secure access to an abortion.

Third, abortion's centrality in politics and political rhetoric makes it a useful demonstration of discordant benevolence during a time of deep partisan division. Abortion is a long-standing focal point in U.S. political life, with opinion on abortion now neatly aligned with partisanship and ideological position (58) . Although Roe v. Wade made abortions legal federally in 1973, abortion continues to motivate political action: It is a litmus test for political candidates, elected officials, and judicial nominees (70) . In the first half of 2021, 90 abortion restrictions were passed in state legislatures, a number greater than the yearly total of any previous year (71) . Abortion has even entered into debates about how to respond to the coronavirus (72) .

In sum, abortion attitudes expose value divergence; getting an abortion commonly necessitates help; and the politicization of abortion makes it a potent arena for value conflict within personal networks. When people decide to have an abortion, they may interact with multiple actors who either proffer or deny help. As with any request to help a friend or family member, norms guiding support provision provide a strong motivation toward prosocial behavior. Helping someone get an abortion, however, might implicate supporters in an act they find morally repugnant. Examining how individuals with different value orientations respond to this dilemma provides an ideal lens through which to understand conflicting values in moral decision-making.

In what proceeds, we address the following empirical research questions: First, how are abortion attitudes associated with people's willingness to offer different kinds of help to friends and family members pursuing an abortion? Second, among individuals who are willing to help in the face of conflicting abortion attitudes, how do they explain their discordant benevolence?

To answer these questions, we take a mixed-methods approach. We draw upon survey data from the 2018 GSS as well as interview data from the largest interview study to date about Americans' abortion attitudes. Both datasets provide insight into Americans' willingness to help someone they know who decides to have an abortion. Understanding people's willingness to help in this scenario furthers our awareness of the on-the-ground reality of obtaining an abortion. These data also enable us to assess the theoretical model of moral decision-making by observing behavioral intentions in the face of conflicting values. The willingness to help in ways that go against one's personal values-what we call discordant benevolencepromises applicability for a wide array of help-seeking scenarios, well beyond the specific case of abortion.

An overwhelming majority of U.S. resident adults are willing to extend help to a close friend or family member who is seeking an abortion. Figure 1 presents the weighted responses to each of the abortion helping questions, as well as the percentage of respondents who would offer at least one form of help. Most Americans would help in a nonpecuniary manner: 88% would provide emotional support and 72% would help with arrangements like a ride or childcare. Over half would help pay for ancillary costs, and around a quarter would help pay for the abortion itself.

The help that Americans are willing to extend is patterned by their stance on abortion morality, as seen in Fig. 2 ; yet, across attitudes on abortion morality, Americans extend support. Those who deem abortion immoral are willing to offer the fewest forms of support and at the lowest levels. Nonetheless, almost half of those who are morally opposed would help a friend or family member with arrangements, and over a third would help with associated costs. A majority of Americans who think the morality of abortion "depends" would offer logistical help and help with associated costs of abortion; just under a quarter would help pay for the abortion itself. These bivariate results are similar when we assess abortion attitudes measured by attitudes toward legality. The results also hold when we instead use a general measure of political ideology (see figs. S1 and S2).

Do Americans with different stances toward abortion's morality still differ on their willingness to offer help when accounting for their ideology, their other abortion opinions (the Rossi scale and insurance coverage), as well as personal characteristics? To answer this, we analyze four logistic regressions predicting whether an individual would offer each type of help. Our strategy takes the same form for each item: We model the relationship between help offered and demographics, ideology, and the full range of abortion attitudes. The full results from the four models are presented in table S2, while With the exception of emotional support, individuals' abortion attitudes are highly predictive of their willingness to offer different types of help. Once we know individuals' attitudes on this set of abortion questions, information on important personal characteristics like age, race, education, religion, and family income do little to improve our ability to predict the forms of help they would offer to a close friend or family member who had decided to have an abortion. Figure 3 shows how an individual's position on abortion morality is associated with their willingness to help, net of demographic attributes, ideological position, and other abortion opinions. Each panel visualizes predicted probabilities for each form of help. Willingness to provide emotional support does not differ by one's stance on abortion morality: Respondents morally opposed to abortion are no less willing to extend this type of help.

Opinion on abortion morality matters greatly, however, for the more instrumental forms of support. Those who are morally opposed to abortion are much less likely than those who say abortion's morality depends or who are unopposed to be willing to help with logistical support, with associated costs, or with paying for the abortion itself.

The differences between people's willingness to extend these three forms of instrumental help are not consistently associated with differences in demographic characteristics such as age, race, religion, or religious attendance, among others (see table S3 ). Rather, their willingness is undergirded by their views: their political ideology, their opinion on abortion legality and insurance coverage, and, most importantly, their opinion on abortion's morality. Opinions on abortion morality are highly predictive of willingness to offer help. At the same time, a substantial number of respondents who are morally opposed or morally equivocal, net of other differences, would nonetheless offer help.

Individuals' values, in particular their stance on abortion morality, are more influential in the decision of how to help than demographic characteristics. Here, we see marked rates of discordant benevolence among Americans who are morally opposed to abortion and are willing to help someone with whom they are close seek an abortion. This holds true even for instrumental forms of support: logistical and pecuniary support. To understand how people make sense of their choices, we turn to the interview data, with a focus on Americans who engage (or imagine they would engage) in discordant benevolence.

How do people willing to extend help despite their moral opposition understand that behavior? What enables discordant benevolence toward a friend or family member who has decided to have an abortion? Here, we explore interviews with Americans who express moral opposition to abortion to understand how they make sense of their discordant benevolence. Rationales for helping (the why) correlate with types of help offered (the how), whether through assistance with arrangements, financial support, emotional support, or otherwise.

Our analyses of qualitative data suggest that interviewees engage in discordant benevolence in one of three ways: as commiseration, as exemption, or as discretion. These three logics illuminate how people make sense of why, despite espousing moral opposition to abortion, they would be willing to help a friend or family member who decides to have an abortion. We anticipate that these logics would not be as prominent in a laboratory setting or interactional settings involving a stranger; they are specific to interactions within established relationships. We also anticipate that they extend beyond the case of abortion.

The first rationale underlying morally opposed interviewees' willingness to help a friend or family member who has decided to have an abortion is commiseration. Commiseration invokes a view of others-especially friends and family-as humans worthy of care despite imperfections. They, like all of us, live in a troubled world and are vulnerable to hardship, adversity, bad luck, and oppression. Casting the woman seeking an abortion within this context of misfortune, commiseration enables morally opposed interviewees to help someone whose choice they disagree with, perhaps adamantly.

Commiseration explains why Maxine (conservative Republican) once drove a friend she met through her addiction recovery group to an abortion clinic. Although adamantly against abortion ("I do believe it is taking life. It is murder."), her friend's experience gave Maxine "a more human view." She describes how "[j]ust seeing how [her friend] was raised and all the things that had happened to her, I guess it gave me more of a viewpoint where, I would still say it's wrong, but I would never tell anyone 'You did wrong,' or condemn them in my mind." Maxine retains her strong "pro-life" identification and wishes that all who are opposed to abortion would "spend their time helping people, you know?"

Lived experience through relationships can enable discordant benevolence via commiseration. Maria (moderate Independent) describes how this happened for her through a teenage friendship:

I had a friend in high school … and she had a very verbally abusive stepfather. And she told me, she was like, "He's going to kill me if he finds out I'm pregnant." She's like, "I'm going to have an abortion," and I told her, I said, "No, don't have an abortion, you know, give the baby to me, I'll raise it." She's like, "That's not the point." She said, "If he finds out I'm pregnant, I'm dead." Maria points to this experience and the commiseration it engendered as the reason why she would be willing to extend help, because "being 'pro-life' is also being pro-women."

Personal experience with an unplanned pregnancy leads other morally opposed interviewees toward helping via commiseration. Roxanne (conservative Republican) recalls how her own abortion experience left her "by myself. I didn't have any support-the boyfriend, friends, anybody." Although she thinks of abortion as "killing," Roxanne says that she would support a woman "who might be in a similar situation," because "we make bad choices, but we're still good people." When Cindy (liberal Democrat) got pregnant "out of wedlock," she recalls being "scared" because "I didn't have anybody." She explains that she is morally opposed to abortion, yet when it comes to others considering an abortion decision, "You can always say, 'I'm here for you'…'I'm here no matter what,'" because "everybody needs somebody to help them through it."

For other interviewees, their willingness to help via a logic of commiseration stemmed from imagining oneself in the situation. This kind of empathy led Alexis (conservative Independent), who labels her own stance on abortion as "pro-support," to say she would help a friend or family member:

If I were ever in that situation, I'd be mortified, and terrified. I'd probably, in my mind, want to get an abortion, but I probably wouldn't, just because-I mean, it was my decision to have sex, unless it was, in whatever case-rape. … And even with some of my friends, too, I would be with them if they chose to get an abortion. Or they chose-were debating about it. I'd let them know I'm there to support them, and be with them, and pray with them, and just love on them. Because that's just such a hard decision to make. And hard situation to be in, in the first place. So, I can only imagine what some people are going through and thinking about.

Alexis concludes that, too often, "the person that's actually going through [it is] just kind of left there in the middle, with no one helping them." The perception that women who decide to have an abortion lack help facilitates help-as-commiseration among morally opposed interviewees.

Correspondingly, help-givers engaging a logic of commiseration may limit what forms of help they offer due to perceptions that their friend or family member will endure pain and emotional difficulty post-abortion [countering this perception, a National Academy of Sciences report summarizes the long-term health effects of abortion as follows: "Based on research that meets scientific standards for rigor and lack of bias, the committee concludes that having an abortion does not increase a woman's risk of secondary infertility, pregnancy-related hypertensive disorders, abnormal placentation (after a D&E abortion), preterm birth (<37 weeks), or breast cancer. Having an abortion also does not increase a woman's risk of depression, anxiety, and/or posttraumatic stress disorder" (73) . Nevertheless, many Americans, including abortion patients, believe that abortion is harmful to women (73) (74) (75) ].

Johnathan (conservative Republican), for example, explains that "if somebody came and told me [they were seeking an abortion], there's always an element of sympathy and empathy and understanding that needs to be given, because the woman who has the abortion suffers." Johnathan does not know of any friends or family members who have experienced an abortion, but roots his willingness to be supportive as commiseration for people who are in "broken" relationships, in poverty, "embarrassed," or who "don't want to deal with pregnancy."

Moral opposition to abortion may deem particular forms of help off-limits. Rick (conservative Republican), for example, says "If you're in need of help, I will do my best to help you," but "don't use my money to pay for [an abortion]." Still other interviewees expressed a commiseration rationale using phrases like "needing mercy," "a spirit of compassion," or "You're still a human being. You're still one of God's children, so I can love you." Such expressions of help were often rooted in hypothetical rather than lived examples of interactions with friends and family members seeking an abortion.

In sum, the logic behind commiseration help is that abortion conflicts with my personal values, but people are imperfect living in an imperfect world, and, therefore, it is okay for me to help. The occasion of a friend or family member making an abortion decision activates empathy within people otherwise unwilling to help at all. Morally opposed interviewees exercising the logic of commiseration hedge that their extension of benevolence (lived or hypothetical) does not mean that they support abortion, nor their friend or family member's decision to abort. A commiseration rationale often prioritizes emotional support (as defined by the help-giver) for the person seeking an abortion and makes allowance for logistical support, but rarely includes financial support. Exemption A second logic underlying interviewees' willingness to help a friend or family member despite moral opposition to abortion is exemption. An extension of help through this rationale acknowledges and retains one's personal opposition to abortion, but carves out a condition of exceptionality to help reserved exclusively for their friend(s) or family member(s). Abortion is a moral "wrong," categorically, but interviewees will (or did) help a specific person in their network (and "only" them) obtain an abortion. This rationale incorporates a view of abortion as morally unacceptable, but enacted alongside the potentially overriding value of helping close others in need.

Ryan (liberal Democrat), for example, describes a moral opposition to abortion that stems from his assessment that a person should "step up to the plate and say, 'You know what, it was my choice, I have to deal with this, and I created this life inside of me and now I have to deal with it.'" But when contemplating the hypothetical scenario of his own sister deciding to have an abortion, Ryan raises an exemption in his willingness to help:

If it were my sister…I would want to talk to her to make sure she's thinking about every possible thing. But if, ultimately, she's like "No, [Ryan], I can handle this," then, "Ok, do what you gotta' do," you know? But it's just because you love someone.

Ryan delimits his sister's hypothetical abortion decision as a special circumstance in which he would be willing to provide help, countering that of a stranger, when "you don't really care." But he draws the line at paying for anyone's abortion: "If we're gonna say that it's the woman's choice, then her choices are gonna come with her own financial assets, as well."

Similarly, Ken (conservative Republican) recounts his willingness to help a girlfriend get an abortion despite moral opposition when "there was an instance when she thought she was pregnant." Ken says they did "a lot of praying for the wrong thing," wishing for a miscarriage to avoid an abortion decision.

A close relationship is what makes helping behavior permissible, via exemption, when help runs counter to another value. Ellie (conservative Republican) says that if it were her own daughter, "I would do everything in my power to say 'I think you're making a bad decision'," but at the same time, "I would still love her as my daughter and support her and do whatever I would need to do." Anne (conservative Republican) hedges on her otherwise staunch moral opposition when contemplating a hypothetical situation in which her granddaughter is pregnant with a baby determined to have a severe health issue. Anne says it would be "very hard" to tell her granddaughter to continue a pregnancy, because she'd want to support her.

Both Anne and Ellie rationalize discordant benevolence as a way to value family foremost.

Unlike those who help through the logic of commiseration, the logic of exemption does not require empathy to enable help-giving. The relationship itself (and unique allowance for help, accordingly) is enough. Interviewees frame their "support" through the promise of continued inclusion. An exemption to "help" kin maintains one's moral disapproval while attempting to offset the potential loss of a relationship. A promise not to end a relationship may or may not be perceived as helpful by the friend or family member, but morally opposed interviewees commonly narrate their support in this way.

The logic behind exemption help, in sum, is that abortion is morally wrong, but this is my friend or family member and so, in this situation, for my friend or family member, I will help. The closeness of the social relationship-namely, that the request for help comes from a friend or family member-explains why they would help at all. An exemption rationale does not compel a change to one's moral positioning on abortion; neither does it imply acceptance of a friend or family member's personal abortion decision. Rather than tying morality to specific abortion circumstances, as in the cases of those who enact discordant benevolence as commiseration, an exemption rationale for help-giving limits benevolence to a specific group of people.

An exemption rationale underscores the importance of role relations in moral decision-making, wherein interviewees emphasize their response to a moral quandary as predicated upon the social relationship. This logic justifies offers of emotional and logistical support, if not financial support, to the friend or family member in question. Close relationships enable discordant benevolence.

The final reason why interviewees with moral opposition to abortion express a willingness to help their friend or family member is that they grant them discretion, or the latitude to make an autonomous personal decision with which the interviewee him or herself disagrees. Discretion differs from both commiseration and exemption in that it makes room for moral individualism (76) , or the idea that right and wrong is a matter of personal opinion. A discretion rationale rests not upon empathy for someone making a "bad" choice, or on treating family and friends as exemptions, but on the perception of a boundary between oneself and a loved one's decision. Abortion can be morally wrong "for me" (the help-giver), but the decision belongs to you (the help-recipient), a distinction that enables helping behaviors.

We heard the logic of help-as-discretion from Lydia (liberal Democrat) who narrates her opposition to abortion by saying, "I'm a mother. I'm a grandmother. I'm a teacher. Children are everything to me. I know so many adoptive parents. What a gift." Lydia believes that "nobody should get an abortion, other than the 'obvious,'" referring to situations of incest, rape, or health. But when Lydia's unmarried son's partner got pregnant unexpectedly, she immediately reached out to help. As Lydia describes: I said, "Okay, well, I need to go see her, and let's have lunch," and I said, you know, I said, "I'm Grandma. Like it or not. I'm Grandma. I'm here. I'm in." I said, "I will help you with whatever you need with this baby. I will pay for an abortion. I will help you through adoption. It's not about me. Just please know that I'm here to help you with whatever you decide to do." Lydia admits being "relieved" when her son's partner instead continued the pregnancy, but affirmed her unqualified assistance when she said, "But would I have paid for an abortion? Yep!" Lydia "would have done whatever [her son's partner] wanted," including traveling to a place with fewer legal restrictions: "We'll just go on over to Germany for a month. I can take you somewhere." Lydia explained her discordant benevolence through the logic of discretion when she said, "I don't think you get to ask all these questions. I think you just say, 'Are you sure?' And, 'Can I give you some brochures? Do you want some counseling?'" Similarly, Melanie (conservative Democrat) describes how her willingness to help a loved one comes from juxtaposed personal values that emphasize an individual's "right to make their own decision" alongside her moral opposition to abortion. Helping college women through abortion experiences reified this for Melanie: I have seen it. When you experience it with someone, it's eye opening. It's a whole different experience. Again, I'm not there to judge. I'm there to provide guidance and support for whatever the decision is. …My heart goes out to someone who is experiencing or questioning an abortion. …My first inclination is that it is wrong. And then I have this other thing sitting out here that says is it a women's prerogative.

Discretion motivates help in the face of personal moral opposition to abortion.

In another example, Veronica (liberal Independent) recalls a time when she drove a best friend to Planned Parenthood to get an abortion. At the time, the friend told her she was getting a cyst removed. Veronica "knew that they didn't do that there, but I didn't question. I just, like, drove out, took her, like, okay, whatever." Months later, Veronica confronted her friend about it, saying, "They don't do that there. I know what they do there," and the friend disclosed her abortion. Veronica says she understands that her friend hid it because she knew Veronica opposed abortion. Discretion motivated Veronica to help, anyway: "For anybody in my life, I'll encourage them to keep [the baby] and I'll break down every way possible for them to make it work. But if they make that choice, there's nothing I can do about it." As with all logics of discordant benevolence, discretion works as a strategy to maintain a close relationship while also maintaining one's own moral stance against abortion.

The logic of discretion is, in sum, that I find abortion morally objectionable for myself, but this is my friend or family member's own independent moral decision to make. My role as a member of their support network is to help them, whatever they decide. To grant discretion is to distance oneself from moral culpability, resonant with libertarianism. Discretion allows for more expansive forms of help including emotional support, paying for the abortion, paying for ancillary costs, helping with arrangements pertaining directly to the abortion procedure, and more. Help-as-discretion enables helpgivers to morally oppose abortion, whether in whole or part, while still supporting their friend or family member who has decided to have an abortion.

Requests for help from friends and family members occasion discordant benevolence. All three logics (commiseration, exemption, and discretion) provide explanations for why interviewees who are morally opposed to abortion said that they have helped or would be willing to help a friend or family member who decided to have an abortion. Discordant benevolence maintains a close relationship alongside a personal abortion stance that counters an abortion-seeker's decision to abort. Table 1 summarizes the characteristics of each of these three logics.

All three rationales facilitate discordant benevolence, providing an answer to the question of "what can I do for my friend/family member who has decided to do something with which I disagree?" Individuals who engage a commiseration rationale enact empathy for the person involved. Persons who engage exemption link their willingness to help to only this help-recipient. Those who engage a discretion rationale make room for autonomous moral discernment.

Across all examples of discordant benevolence, the meaning of "help" is reframed through help-givers' own values. Help is not offered in equal measure, but circumscribed by help-givers as a 

When Americans who are morally opposed to abortion are asked to help a close friend or family member obtain one, they face a moral crossroads. Do they support those with whom they live in community or enact their moral opposition to abortion? Here, we ask: How do people manage this fraught decision? Drawing upon both quantitative and qualitative data, we find that often Americans do not choose one path over another. Rather, they engage in discordant benevolence. They extend support but discriminate among forms of help.

Americans are more willing to extend emotional support or to assist with the logistics of a close friend or family member's abortion than they are to help finance the procedure or its ancillary costs, as revealed through the GSS data. This distinction may reflect the social meaning of money (77) , whereby spending money is a way to enact one's values. Refusing to contribute directly to the procedure may be a strategy people who are morally opposed to abortion use to mitigate their conflicting values, putting acceptable distance between their help and the abortion itself. Money is fungible, however, so this action is a form of symbolic resistance (i.e., funds contributed to ancillary costs can easily be redistributed to the procedure). The finding that people discriminate among types of help, moreover, is consistent with research suggesting that medical professionals who assist in abortion care draw distinctions between actions that constitute participating in abortion procedures versus those that involve caring for the patient before or after (78) . Professionals use circumscribed medical care to resolve conflict between their opposition to abortion and their sense of obligation to provide care to all persons (79) .

Americans discursively make sense of discordant benevolence by offering three different, and sometimes overlapping, explanations: commiseration, exemption, and discretion. Each differs in the way in which morally conflicted help-givers cast the person in need, and each correlates with different types of help. With commiseration, those seeking help are imperfect and mistaken, buffeted by the misfortunes of a troubled world. This makes them worthy of support. With exemption, the close relationship justifies the unusual extension of help. Both commiseration and exemption make room to help according to help-givers' own framing of who needs and what is needed. Discretion permits morally opposed help-givers to separate their personal moral opposition from an abortion-seeker's own moral decision-making. This logic is most permissive of forms of help, whether as emotional support, financial help, or help with logistical arrangements for an abortion.

While scholars have long noted that people hold inconsistent beliefs (80) , we argue that what may appear inconsistent here is not. Many Americans' abortion opinion is informed by genuine internal value conflict (8) . Since at least the 1980s, a subset of Americans say that abortion should be legal but are morally unsure or morally opposed to abortion themselves [for example, (80) ]. Furthermore, abortion opinion may result from conflict between values of individual autonomy and traditional gender roles (54) . Last, internal divisions reflect the way abortion is dealt with by the law: legal in some circumstances, illegal in others. Similarly, internal conflict between two deeply held values-supporting those within one's social network and abortion opposition-results in discordant benevolence.

Examining instances of conflicting values arising from interpersonal requests for support reveals important implications for foundational sociological theories. First, we extend the moral decision-making literature by attending to the dyad, uncovering the role of benevolence in motivating prosocial behavior within close personal relationships. Much of the literature on moral decision-making emphasizes individuals' actions toward abstract others, such as the beneficiaries of charitable giving, largely ignoring the interpersonal and network context. Most benevolence, however, occurs within existing relationships. Examining moral decision-making within personal networks enables us to demonstrate two things: the practice of discordant benevolence and the logics by which people make sense of it. Examining only behavior toward strangers has obscured the importance of normative pressures governing close social ties in moral decisionmaking. We demonstrate that social ties are critical, motivating even Americans ardently opposed to abortion to extend help to a friend or family member who has decided to get an abortion. The highly salient value of benevolence in the context of a close relationship may be nearly impossible to eschew. Our survey data reveal a strategy of negotiating these conflicting moral constraints by circumscribing helping behavior to reduce dissonance caused by competing personal values. Our interview data reveal the ways in which people reason through their response to these conflicts.

Sociological insights regarding the role of relationships and social context in moral decision-making are largely absent in the sociology of morality (40) , but our work illustrates their centrality. People's judgments about the morality of a given action are strongly dependent upon the characteristics of the actors involved, with exemptions and commiseration disproportionately granted to actors that are otherwise familiar, valued, or held in high social esteem. Examining prosocial behavior or attitudes in isolation permits "rational," cognitive decision-making, but obscures the almost taken-for-granted inclination to support and comfort loved ones in crisis. Only when we examine such decisions in the context of actual or "imagined relationships" (4) do we reveal the blurry moral lines and emotion-laden processes that enable discordant benevolence.

Second, we build on long-standing debates surrounding the role of motivation in action. Our findings provide insight into how people cope when at a moral crossroads, a "problematic situation" (81) . We find that people resolve cognitive dissonance by offering circumscribed help that nonetheless holds a moral line. That is, we find that support offered is patterned by the strength of moral opposition: Those who are the most morally opposed are willing to offer fewer and less instrumental forms of help than those who do not hold as firm a moral opposition. The interview data reveal how Americans explain moral decision-making in retrospect or hypothetically by selectively (de)emphasizing or reprioritizing different values to justify actions that are value-discordant (or seem so on their face). Thus, moral conflict is negotiated through cultural, cognitive, or behavioral adaptation. In marshaling multiple kinds of evidence, we reveal "how moral variability between persons interacts with the moral meanings of situations to influence human conduct" (3) and further illuminate moral systems as neither monolithic nor necessarily coherent.

Third, we extend theories of social support by identifying personal values as a critical component of support provision and help-seeking processes. To date, the literature has overlooked the role of personal values in mitigating the extension of support to friends and family. We find that personal values intervene in support processes in complex ways, influencing decisions not only about whether to provide support but also about the nature and extent of support that people are willing to provide.

Fourth, our findings also have implications for personal interaction amid political division. That Americans are ever more politically polarized is a familiar chorus, and the issue of abortion is central to it. Disparate attitudes on abortion's morality and legality reflect ideological division observed elsewhere (7, 70) . Given this narrative, we might expect abortion attitudes to be extremely hard to overcome in decisions about extending help to others. However, we demonstrate that moral and political absolutes are often subjugated by opportunities for prosocial action on behalf of friends and family. People are willing to cross ideological and partisan lines to help others within their personal networks. Thirty percent of Americans who are morally opposed to abortion are still willing to spend their money to help a friend or family member get one; 45% will help with logistics and arrangements. This does not reflect an inconsistency in abortion values, but rather the salience of a competing value: benevolence. This transcendence of personal ties over politics in one of the most polarizing issues in American lifeabortion-illustrates the fundamental human drive to maintain connectedness and the strength of obligation to in-group members.

This mixed-method analysis was not envisaged until both sets of data were collected, introducing some weaknesses to the evidence we can bring to bear. Ideally, the qualitative and quantitative data collection efforts would inform one another concurrently. The interview protocol did not ask directly about providing help in obtaining an abortion. Rather, these narratives emerged in the context of other prompts. However, interviewees were asked the GSS abortion morality questions as well as other abortion attitude questions and standard survey questions. This permitted us to align the interviewees with respondents from the GSS, leveraging different kinds of data to examine similar individuals. A strength of executing the data collection efforts separately is that we achieved independent confirmation from the interviewees regarding the importance of helping as an axis for understanding abortion in the United States and the presence of discordant benevolence.

In addition, because the two datasets were collected independently, we cannot ensure that the types of help identified in the GSS survey are interpreted similarly in the interview data. We make inferences as to the variety of meanings interviewees attach to the term "help." This is most concerning with regard to the category of emotional support. The quantitative findings show high rates of willingness to provide emotional support across the spectrum of attitudes toward abortion morality as well as in each of our sensitivity analyses (looking at abortion legality attitudes and general political ideology). While the other examples of help in the GSS are concrete forms of extending and receiving support (a ride, childcare, money for a hotel, money for the abortion itself), emotional support is not as clear-cut. Emotional support can span from passively listening to offering advice and affirmation. Furthermore, emotional support can be welcome or unwelcome if perceived as patronizing or intruding on one's privacy or autonomy. Variation in emotional support and its reception make it more complicated than, for example, offering and accepting money. Concerns about how to interpret emotional support do not threaten our theoretical argument, however, because we observe discordant benevolence within the categories of instrumental help as well.

Our data provide both hypothetical and lived examples of help, but we rely heavily on people's articulated willingness to provide help. We stand by this approach given prior work showing how forcedchoice surveys on hypothetical situations predict behavior by revealing unconscious decision-making (5) .

Our focus on the help-giver-as opposed to the help-seekerraises questions that constitute important avenues for future research. Here, we offer insights into the experience of value conflict from the exclusive perspective of the help-giver. What would we learn from examining both sides of the dyad in discordant benevolence: the help-giver and the help-seeker? We could learn, for instance, within what kinds of relationships and networks the moral obligation to support others is the dominant value. Building upon what we have done here, answering this question would bring further insights to both the moral decision-making literature and the support within social networks literature.

We further encourage researchers to examine logics and offers of help among people across the abortion attitude spectrum. As the first study to consider value conflicts within helping requests, we targeted our analysis to individuals who faced the greatest discordance but nonetheless agreed to help. This means that, in the interview data, we did not examine people who were not willing to offer help or those whose values were less conflicted.

The fundamental question of this research is how do people manage when at a moral crossroads? The answer is that some people engage in discordant benevolence. That is, they are willing to help friends or family pursue actions they find disagreeable or even reprehensible. How do we understand this behavior? A facile response is that people are hypocritical. We see it otherwise: Individuals face conflicting values, all of which are sincerely held. Americans morally opposed to abortion who are willing to help someone they care about obtain an abortion often make sense of this resolution by circumscribing the types of help they proffer and by using logics of commiseration, exemption, and/or discretion to explain this resolution.

The phenomenon of discordant benevolence extends beyond abortion into other morally fraught arenas, such as divorce (82) and euthanasia (8) . We contend that discordant benevolence is a central feature of interactional life within social networks. Our major contribution has been to document these actions in the context of settings consequential to people's lives and to illustrate ways people make sense of their discordant benevolence.

We marshal both large-scale quantitative and qualitative data to investigate discordant benevolence. The quantitative data reveal the presence, frequency, and correlates of people's willingness to help someone they know do something they may find morally objectionable, in this case seek an abortion. These data also demonstrate how values are correlated with the types of help people are willing to extend. The qualitative data reveal how people extending discordant benevolence make sense of their resolution to the moral crossroads.

We use data from the 2018 GSS for our quantitative analysis. The GSS has been fielded by the National Opinion Research Center at the University of Chicago since 1972, with support from the National Science Foundation. The GSS uses multistage, full-probability sampling designed to be representative of English-and Spanish-speaking adults living in U.S. households. In 2018, the GSS interviewed 2348 people, 59% of those sampled. Among the interviews, 92% were in person and 8% were done by telephone. In 2018, the GSS was fielded across three ballots; analyses conducted here use ballots A and C (N = 1574) because the set of abortion opinion measures outlined below were asked only in these two ballots.

The module examining provision of help for a person obtaining an abortion appeared in the GSS for the first time in 2018. It reads:

"If a close family member or friend decided to have an abortion, which of the following kinds of help, if any, would you give if you were able… Help with arrangements, like a ride or childcare? Help paying for the abortion? Help paying for costs other than the abortion, like for a ride or hotel if she needs to stay overnight?

Help by providing emotional support?" Respondents were given response options of "yes" or "no." In our sensitivity analysis, we summed positive responses to the set of helping questions to produce a "Help Index." Respondents who volunteered "don't know" or who did not answer one or more items were dropped from the analysis corresponding to that helping item (between 3 and 5% of respondents asked).

Our key independent variable of interest is an item that asks about the morality of abortion, also fielded for the first time in the 2018 GSS. The question reads: "Leaving aside whether you think abortion should be legal, are you morally opposed to abortion or not, or would you say it depends?"

To attend to respondents' opinions on other aspects of abortion attitudes, we use the traditional Rossi scale as well as new abortion questions fielded for the first time in 2018, outlined below. The Rossi scale measures support for legal abortion on a six-point scale, where 0 denotes a respondent who thinks a pregnant woman should not be able to obtain a legal abortion under any of the hypothetical circumstances asked, and 6 denotes a respondent who thinks a pregnant woman should be able to obtain a legal abortion in all of the hypothetical circumstances. The scale was produced by summing positive responses to each of the six constituent items; hence, individuals who responded "don't know" to one or more items were dropped (11% of respondents asked). The six constituent questions have the following wording: "Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion...a) if there is a strong chance of serious defect in the baby? b) if she is married and does not want any more children? c) if the woman's own health is seriously endangered by the pregnancy? d) if the family has a very low income and cannot afford any more children? (e) if she became pregnant as a result of rape? (f) if she is not married and does not want to marry the man?" A seventh question, introduced in 1977, asks "if the woman wants it for any reason?" In theory, this question ought to receive fewer endorsements than any of the others, and no one who said "no" to one of the specific questions should say "yes" to the encompassing one. Our analyses show that this is not the case: 20% of respondents who thought a woman should not be able to obtain a legal abortion in at least one of the hypothetical circumstances answered "yes" to "any reason." We thus omit this question from our scale. As a sensitivity analysis, we also ran our models with this question included in the scale and found no substantial differences.

In addition to the Rossi scale items, we also analyze the following questions on health insurance coverage, newly fielded in the 2018 GSS: "People use their health insurance to help cover the cost of receiving health care. Do you think people should be able to use their health insurance to help cover the cost of receiving an abortion?" Response options were binary: "people should be able" and "people should not be able."

We checked for multicollinearity between the items measuring abortion opinion in all regression analyses with more than one abortion item as an independent variable. Variance inflation factors were below 3. We chose to use the Rossi scale as opposed to new item on abortion legality given that this scale has informed public opinion research since the 1970s; our results are the same when the Rossi scale is replaced by the legality item.

In addition to items measuring abortion opinion, we use a standard set of demographic and ideological controls as potential correlates with abortion public opinion (83) and prosocial behavior (84) . These include sex (1 = female, 0 = male), age, race (white, black, Hispanic or Latino/a, all other), marital status (married, widowed, divorced, separated, never married), employment status (full-time, part-time, unemployed, retired, keeping house, in school, or other), religion (Conservative Protestant, Mainline Protestant, Catholic, Jewish, other, no religion), religious attendance (1 = weekly attendance, 0 = less than weekly attendance), education (no high school diploma, high school graduate, some college, BA or higher), family income (1 = earning $90,000 or more, 0 = earning less than $90,000), region (Northeast, Midwest, South, and West), U.S. residency at age 16 (1 = U.S. resident at age 16, 0 = not a U.S. resident at age 16), and ideology (self-reported views on a seven-point scale from "extremely liberal" to "extremely conservative"). Our coding of religious affiliation is a modified version of the RELTRAD classification (85) , with the category of Black Protestant folded into the category of Conservative Protestant. As Schnabel (86) notes, this avoids conflating religion and race. Our family income measure is coded to split respondents by average income. The coding of these covariates balances model parsimony given our relatively small sample sizes with the aim of splitting respondents into meaningful subgroups. However, many different iterations of model specifications were run, for example, with age treated categorically, religious attendance treated categorically, and income treated continuously. Our results are robust to these model choices. Given recent findings that marginalized groups are, on average, simultaneously more liberal and more religious (87), we ran models with interactions between race and gender. None of these interactions were statistically significant, nor did they alter substantively the interpretation of other variables in the models.

Between March and August 2019, a team of five researchers including two of the authors conducted a total of 217 in-depth interviews across six regions in the United States for the NAAS. The study set out to learn more about how and what "ordinary" Americans, as opposed to only activists, think about abortion.

The NAAS qualitative interview sample strategically combined a random probability sample and nonprobability quota sample. Following an initial pilot (n = 20), we recruited a pool of potential interviewees using a 2500-piece, random, address-based mailing to zip codes within six states: California, Colorado, Indiana, North Dakota, Pennsylvania, and Tennessee. The recruitment letter continued to next page invited recipients to complete an online pre-screener to potentially qualify for a paid interview about an unspecified "social issue" (see the Supplementary Materials). Abortion was not revealed as the topic until the final question of the online pre-screener. The online pre-screener gathered interviewees' key demographics as well as their political orientation, religious affiliation and attendance, and ideology (using the GSS seven-point scale). Using these responses, a sample was constructed to approximate the U.S. distri bution of characteristics most closely associated with Americans' attitudes toward abortion. The sample was also balanced by gender, race, and age.

In some geographic zones, meeting quota goals meant implementing supplemental recruitment beyond the initial mailing. In two geographic zones, a follow-up postcard was mailed 6 weeks after the initial mailing to boost response rates. Additional strategies entailed a mix of targeted snowballing, flyer distribution in neighborhoods/ organizations exhibiting underrepresented quota characteristics, and, in one zone, an online advertisement recruiting "moderates" (still without disclosing abortion as the topic). Across all strategies, responses to the online pre-screener (total N = 671) determined inclusion according to quota needs.

The resulting NAAS sample (n = 217) includes 72 "liberals" (1 to 3 on the ideology scale), 72 "moderates" (4 on the ideology scale), and 73 "conservatives" (5 to 7 on the ideology scale). A subset of interviewees (n = 74) constitutes the focus of this paper (see the "Analysis of interview data" section).

Interviews were conducted in person at a semiprivate location, typically a library. Exceptions for telephone interviews were permitted if interviewees declined to meet in person. Interviews were confidential, audio-recorded, and lasted approximately 75 min each. Researchers took note of setting, participant characteristics, and non-audible expressions. Interviewees received $30 for their participation. All interviews were transcribed, and interviewees were assigned pseudonyms to ensure confidentiality.

The bulk of the semistructured NAAS interview protocol consisted of original questions designed to elicit (i) attitudes toward abortion, (ii) influences on those attitudes (religious, ideological, occupational, philosophical, political, interpersonal, experiential, and more), (iii) moral and legal stances toward abortion, and (iv) engagement with abortion as an issue via interpersonal communications, social media, traditional media, politics, and activism. The interview protocol also replicated several measures from the GSS and Gallup about abortion morality and legality as well as interviewees' selfidentification as "pro-choice" or "pro-life." Responses to these questions facilitated the parallel analysis of NAAS data alongside GSS data for the purposes of this article.

Mixed-methods research designs heighten validity by not only confirming results but also providing an evidentiary basis to explain them. This article takes a complementary approach (88, 89) by combining analyses of two separate studies-GSS and NAASconducted in proximate but nonoverlapping time periods. Findings that emerged from quantitative results of the 2018 GSS were explored for confirmation and explanation in the qualitative results of the 2019 NAAS.

Although conducted separately, the NAAS replicated several demographic, ideology, and abortion-specific attitudinal questions asked in the GSS. NAAS interviewees were asked, verbatim, the full battery of questions from the Rossi scale as well as the aforementioned 2018 GSS question regarding the morality of abortion ("Leaving aside whether you think abortion should be legal, are you morally opposed to abortion or not, or would you say it depends?"). As in the GSS, NAAS interviewees could choose from preset response categories or instead say "I don't know" or refuse to answer. Unlike the GSS, the NAAS allowed interviewees to volunteer a response outside the preset GSS response categories, and NAAS interviewees could explain why they answered each GSS question the way they did. Additional probing used cognitive interviewing techniques (90) , inviting interviewees to clarify their own understanding of survey question terms, share their mental imagery, and draw relevant personal connections.

Asking an identical subset of questions on both the GSS and the NAAS links these two studies-one primarily quantitative and the other primarily qualitative-together for seamless mixed-methods analyses. Both GSS and NAAS participants can be sorted by views on abortion morality, Rossi scores, ideology, key demographics, and more. This means that the analytic samples for the GSS and NAAS mirror one another, although the actual participants differ.

While our work focuses on those engaging in discordant benevolence (that is, those who are morally opposed to abortion but would nonetheless assist someone they know seeking one), we provide quantitative results for the entire range of attitudes toward abortion morality. We explicitly compare those who are morally opposed to abortion to those for whom the morality of abortion depends and those who are not opposed to the morality of abortion.

We use the in-depth interview data to illuminate the rationales of people who engage in discordant benevolence. We exclude from our analysis those interviewees who held that they are not morally opposed to abortion; for this group, a request for help would not invoke conflicting values. We include in our analysis interviewees who are morally opposed but who decline to help, sever the tie, or attempt to dissuade their friend or family member from having the abortion. However, given our focus on the theoretically rich case of discordant benevolence, we report only on those interviewees who are morally opposed and express a willingness to offer help. We leave the other conceptually distinct strategies for future research.

Our quantitative analysis is motivated by the question: How are abortion attitudes associated with people's willingness to offer different kinds of help to friends and family members pursuing an abortion? To this end, we analyze four logistic regressions predicting whether an individual would offer each form of help. Because we are interested in discordant benevolence, when a request for help inspires a moral quandary, the independent variable of primary interest is respondent attitude regarding abortion morality.

In these models, we include respondent demographics, abortion opinion, and political ideology. Our independent variables of interest are a respondent's opinion on abortion morality as well as abortion attitudes captured by the Rossi scale and a question on insurance coverage for abortion. We focus our interpretation on opinion on abortion morality. Of the GSS respondents who were asked the relevant abortion items (n = 1395), 127 gave no answer or responded "don't know" to one or more of the four helping questions. A further 139 gave no answer or said "don't know" to one or more of the Rossi questions. The rest did not give information on the demographic or ideology controls. The models are a complete case analysis: between 1100 and 1157 respondents across the four helping items.

We conducted numerous sensitivity analyses of the GSS data. First, out of a concern that missingness was biasing our results, we determined whether any of the control variables predicted whether a respondent answered the four helping questions (table S2) . Few of the control variables were significant predictors; women compared to men and black respondents compared to white respondents were less likely to answer all four helping items, while those who had never been married were more likely to answer the items compared to those who were married.

Second, we were concerned that any differences between paying for the ancillary costs of the abortion and paying for the abortion itself might be related only to a perception of the level of financial assistance required. These concerns are assuaged by two things: First, the question specifies respondents' willingness to help pay, not pay for it entirely. Second, we examined only the highest earners (families with incomes equal or greater than $90,000)-those for whom we would anticipate financial constraints would be less present-and observed the same patterns as for respondents with all incomes. Hence, we believe our findings to be robust.

The purpose of the qualitative analysis of this paper is to explain how Americans who engage in discordant benevolence make sense of this resolution to their moral quandary. The NAAS interview protocol did not explicitly ask the closed-ended GSS helping questions, but it nonetheless elicited relevant examples, both actual and hypothetical. The semistructured, conversational nature of the NAAS interviews meant that participants had an opportunity to share experiences, tell stories, and express their own connections to the issue of abortion. "Helping" discussions thus arose naturally. A question about interviewees' personal experiences with abortion (whether of the interviewee or of someone they knew), for example, evoked discussions of helping. Queries about recent conversations on the topic likewise produced relevant responses. Researchers probed responses further as a means of eliciting stories, explanations, rationales, and emotions.

The current analysis is targeted specifically to the subset of interviewees who said "morally opposed" in response to the GSS question replicated within NAAS interviews: "Leaving aside whether you think abortion should be legal, are you morally opposed to abortion or not, or would you say it depends?" (n = 74). This subset represents those for whom a request to help obtain an abortion could illicit the most value conflict. Table 2 contains a breakdown of interviewee characteristics for this analytic sample. Transcriptions from this subset of interviewees were read closely and coded to identify examples of emotional, financial, and other forms of help extended to friends and family who had decided to have an abortion. Examples could be actual or hypothetical. "Emotional support" was inferred from expressions of sympathy, empathy, counsel, "support," and related sentiments. Analytic memos put coded excerpts in context, synthesized quotations across transcripts, and ascertained recurrent themes in the qualitative data.

Social relationships and health

An overview of the Schwartz theory of basic values. Online Read

The new sociology of morality

paper presented at the 115th Annual Meeting of the

Motivation and justification: A dual-process model of culture in action

Have American's social attitudes become more polarized?

Was there a culture war? Partisan polarization and secular trends in US public opinion

Hard Choices, Easy Answers: Values, Information, and

Abortion incidence and service availability in the United States

The emergence of human prosociality: Aligning with others through feelings, concerns, and norms

The evolution of human ultrasociality

Normative influences on altruism

Mechanisms linking social ties and support to physical and mental health

Delineating personal support networks

Morality is personal

When a virtue is not a virtue: Conditional virtues in moral evaluation

Asymmetry in judgments of moral blame and praise: The role of perceived metadesires

Uncalculating cooperation is used to signal trustworthiness

Fundamental dimensions of social judgment: Understanding the relations between judgments of competence and warmth

Foundations of cooperation in young children

Beyond altruism: Sociological foundations of cooperation and prosocial behavior

Diversity in social support by role relations: A typology

Brothers' keepers: Situating kinship relations in broader networks of social support

Structures of Social Life: The Four Elementary Forms of Human Relations: Communal Sharing, Authority Ranking, Equality Matching, Market Pricing

Perceived dimensions of interpersonal relations

Someone To Talk To

Different strokes from different folks: Community ties and social support

Prosocial Motives, Emotions, and Behavior: The Better Angels of our Nature

Values, valences, and choice: The influences of values on the perceived attractiveness and choice of alternatives

Internalized values as motivators of altruism, in Development and Maintenance of Prosocial Behavior

Value priorities and behavior: Applying a theory of integrated value systems

Positive Social Behavior and Morality: Social and Personal Influences

Composition of the workplace and psychological well-being: The effects of tokenism on America's black elite

Universalism values and the inclusiveness of our moral universe

The neuroscience of social decision-making

Elicitation of moral obligation and self-sacrificing behavior: An experimental study of volunteering to be a bone marrow donor

Positive and negative referent others as sources of influence: A case of helping

Responsibility, norms, and helping in an emergency

Theorizing moral cognition: Culture in action, situations, and relationships

A theory of the self for the sociology of morality

Emotion and ethical decision-making in organizations

Emotion and decision making

A special emphasis and look at the emotional side of ethical decision-making

The (re)genesis of values

Tools from moral psychology for measuring personal moral culture

Culture and choice: Toward integrating cultural sociology with the judgment and decision-making sciences

The emotional dog and its rational tail: A social intuitionist approach to moral judgment

The new synthesis in moral psychology

Illness careers and network ties: A conceptual model of utilization and compliance

Beyond rational choice: The social dynamics of how people seek help

Measuring stability and change in personal culture using panel data

abortion attitudes remain closely divided

Abortion and the Politics of Motherhood

World view and abortion beliefs: A replication of Luker's implicit hypothesis*

Between Two Absolutes: Public Opinion and the Politics of Abortion

Updating a time-series of survey questions: The case of abortion attitudes in the general social survey

Unstable Majorities: Polarization, Party Sorting, and Political Stalemate

Differences in abortion service delivery in hostile, middle-ground, and supportive states in 2014

Out-of-pocket costs and insurance coverage for abortion in the United States

Barriers to abortion care and their consequences for patients traveling for services: Qualitative findings from two states

Characteristics of U.S. abortion patients in 2014 and changes since

Regulating insurance coverage of abortion

State funding of abortion under Medicaid

Identifying national availability of abortion care and distance from major US cities: Systematic online search

Counseling and waiting periods for abortion

Characteristics of US abortion patients in 2014 and changes since

Secrets and misperceptions: The creation of self-fulfilling illusions

Enacted abortion stigma in the United States

State policy trends at midyear 2021: Already the worst legislative year ever for U.S. abortion rights

Texas and Ohio include abortion as medical procedures that must be delayed

The Safety and Quality of Abortion Care in the United States

Connecting knowledge about abortion and sexual and reproductive health to belief about abortion restrictions: Findings from an online survey

Common misconceptions about abortion risks: What do abortion patients believe?

Lost in Transition: The Dark Side of Emerging Adulthood

The Social Meaning of Money: Pin Money, Paychecks, Poor Relief, and Other Currencies

Conscience reconsidered: The moral work of navigating participation in abortion care on labor and delivery

Willing and Unable: Doctors' Constraints in Abortion Care

The nature of belief systems in mass publics

Later Works of John Dewey

Uncoupling: Turning Points in Intimate Relationships

Gender and abortion attitudes: Religiosity as a suppressor variable

Exploring the structural contexts of the support process: Social networks, social statuses, social support, and psychological distress

Social Networks and Health

The measure of American religion: Toward improving the state of the art

Sexual orientation and social attitudes

Opiate of the masses? Inequality, religion, and political ideology in the United States

How to conduct a mixed methods study: Recent trends in a rapidly growing literature

Mixed methods and the logic of qualitative inference

Cognitive Interviewing: A Tool for Improving Questionnaire Design

Defining generations: Where millennials end and generation Z begins

We thank M. Hout, D. Baldassarri, K. Kimport, P. Tevington, and the Indiana University Department of Sociology's Medical Sociology Working Group for helpful comments on prior drafts. We thank C. Fischer and R. Almeling for sharing their thoughts with us as we developed this work. The reviewers and editors provided exceptionally productive and generative feedback. Funding: Developing the survey questions and fielding them on the GSS was supported by an anonymous private foundation. The interview data collection and analysis was supported by the University of Notre Dame. Author contributions: Conceptualization: S.K.C., T.C.B., and B.P. Methodology: S.K.C., T.C.B., B.L.P., B.R., S.P., and E.M.A. Investigation: T.C.B., B.R., and S.P. Writing: S.K.C., T.C.B., B.L.P., B.R., S.P., and E.M.A.

Supplementary material for this article is available at https://science.org/doi/10.1126/ sciadv.abj5851 Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. The GSS data are available at www.gss.norc. org/. The replication code can be downloaded from the Harvard Dataverse: https://doi. org/10.7910/DVN/V1Y2A5. The interview data are not available per institutional review board protocols. Public availability of interview data would compromise participant privacy. In-depth interviews contain sensitive information pertaining to participants' private and professional lives. Participants did not consent to public availability of their full interview transcripts.